Lightweight MAC protocol that allows for duty cycling to save energy. More...
Lightweight MAC protocol that allows for duty cycling to save energy.
LWMAC adopts the radio duty-cycle scheme to conserve power. Namely, in each cycle period (MAC superframe), a node device wakes up for a short period of time (called listen period or wake-up period) for receiving possible incoming packets from other devices. Outside the listen period, the node device turns off its radio to conserve power.
LWMAC adopts the phase-lock scheme to further reduce power consumption. Each node device in LWMAC will try to record/track its Tx-neighbor's wake-up phase. This is called phase-lock. After phase-locking, the sender node will (likely) spend less preamble packets (also called WR packet, i.e., wake-up-request, in LWMAC) for initiating a hand-shaking procedure for transmitting a data packet, compared to the first time it talks to the receiver.
LWMAC adopts pending-bit technique to enhance its throughput. Namely, in case of having multi packets for the receiver, a sender uses the pending-bit flag embedded in the MAC header to instruct this situation, and the buffered packets will be transmitted in a continuous sequence, back to back, to the receiver in one shot.
LWMAC adopts auto wake-up extension scheme based on timeout (like T-MAC). In short, when a packet is successfully received at the receiver side, the receiver will reset the wake-up timeout to extend its wake-up period for receiving more potential incoming packets. This is to be compatible with the pending-bit technique to allow the receiver to absorb more packets when needed, thus boosts the throughput.
LWMAC adopts a simple retransmission scheme to enhance link reliability. The data packet will only be dropped in case the retransmission counter gets larger than GNRC_LWMAC_MAX_DATA_TX_RETRIES.
LWMAC adopts an automatic phase backoff scheme to reduce WR (preamble) collision probability. In multi-hop scenarios, let's say, nodes A <—B <-—C (which is common in multi-hop data collection networks), in which B has packets for A, and C has packets for B. In case A and B's wake-up phases are too close (overlapping). Then, especially in high traffic conditions, B and C may initiate transmissions at the same time (B sends to A, and C sends to B), a link of either will be definitely interfered, leading to collisions and link throughput reduction. To this end, by using the automatic phase backoff scheme, if a sender finds its receiver's phase is too close to its own phase, it will run a backoff scheme to randomly reselect a new wake-up phase for itself.
|Header definition LWMAC. |
|Interface definition for the LWMAC protocol. |
|Timeout handling of LWMAC. |
|Definition of internal types used by LWMAC. |
|Interface definition for internal functions of LWMAC protocol. |
|Implementation of RX state machine. |
|Implementation of TX state machine. |
|#define||GNRC_LWMAC_WAKEUP_INTERVAL_US (100LU * US_PER_MS)|
|Time between consecutive wake-ups. More...|
|#define||GNRC_LWMAC_PREAMBLE_DURATION_US ((13LU * GNRC_LWMAC_WAKEUP_INTERVAL_US) / 10)|
|The Maximum WR (preamble packet gnrc_lwmac_frame_wr_t) duration time. More...|
|#define||GNRC_LWMAC_TIME_BETWEEN_WR_US (5U * US_PER_MS)|
|Timeout to send the next WR in case no WA has been received during that time. More...|
|#define||GNRC_LWMAC_WAKEUP_DURATION_US (GNRC_LWMAC_TIME_BETWEEN_WR_US * 2)|
|How long a node in LWMAC should keep awake and listen on the channel in one cycle. More...|
|#define||GNRC_LWMAC_BROADCAST_DURATION_US ((GNRC_LWMAC_WAKEUP_INTERVAL_US * 11) / 10)|
|How long broadcast packets gnrc_lwmac_frame_broadcast_t will be sent to make sure every participant has received at least one copy. More...|
|Time to idle between two successive broadcast packets, referenced to the start of the packet. More...|
|#define||GNRC_LWMAC_WR_PREPARATION_US ((3U * US_PER_MS))|
|WR preparation overhead before it can be sent (higher with debugging output). More...|
|#define||GNRC_LWMAC_DATA_DELAY_US (10U * US_PER_MS)|
|How long to wait after a WA for data to come in. More...|
|CSMA retries for DATA packet after WR->WA was successful. More...|
|Maximum TX transmission retries for DATA packet in case of no response from the receiver. More...|
|#define||GNRC_LWMAC_MAX_TX_BURST_PKT_NUM (GNRC_LWMAC_WAKEUP_INTERVAL_US / GNRC_LWMAC_WAKEUP_DURATION_US)|
|MAX burst transmission packet number in one shot. More...|
|MAX bad Listen period extensions a node can tolerate. More...|
|CSMA retries for broadcast packet. More...|
|Default message queue size to use for the LWMAC thread. More...|
|kernel_pid_t||gnrc_lwmac_init (char *stack, int stacksize, char priority, const char *name, gnrc_netdev_t *dev)|
|Initialize an instance of the LWMAC layer. More...|
|#define GNRC_LWMAC_BROADCAST_CSMA_RETRIES (3U)|
|#define GNRC_LWMAC_BROADCAST_DURATION_US ((GNRC_LWMAC_WAKEUP_INTERVAL_US * 11) / 10)|
How long broadcast packets gnrc_lwmac_frame_broadcast_t will be sent to make sure every participant has received at least one copy.
Since LWMAC adopts duty-cycle scheme, a node only wakes up for a short period in each cycle. Thus, when a node wants to broadcast a packet, it repeatedly broadcasts the packet for one GNRC_LWMAC_BROADCAST_DURATION_US duration which is slightly longer than GNRC_LWMAC_WAKEUP_INTERVAL_US. This is to ensure that all neighbors will not miss the broadcast procedure of the sender and catch at least one copy of the broadcast packet.
|#define GNRC_LWMAC_DATA_CSMA_RETRIES (3U)|
CSMA retries for DATA packet after WR->WA was successful.
After receiving the WA packet gnrc_lwmac_frame_wa_t from the receiver, the sender starts sending the data packet using CSMA/CA. This macro defines how many CSMA retries a sender will be allowed to execute for sending its data, before the data is successfully sent (gets data ACK from the receiver).
|#define GNRC_LWMAC_DATA_DELAY_US (10U * US_PER_MS)|
How long to wait after a WA for data to come in.
When a node in LWMAC gets a WR during its wake-up period, it immediately replies a WA packet to the sender for acknowledging the sender's transmission request. After sending the WA, the receiver waits for the data packet from the sender, with a timeout of GNRC_LWMAC_DATA_DELAY_US duration. In case no data will be received in this period, the receiver regards reception failed and go back to normal listen mode. However, in case the receiver receives other unintended packets, like WR/WA packets from other neighbor communication pairs, the receiver resets this timeout and continues to wait for the data packet, with the consideration that the sender's data transmission might be delayed due to other ongoing transmissions (the data packet is transmitted with CSMA/CA). This data timeout is long enough to catch the beginning of the packet if the transceiver supports NETDEV_EVENT_RX_STARTED event (this can be important for big packets).
|#define GNRC_LWMAC_IPC_MSG_QUEUE_SIZE (8U)|
|#define GNRC_LWMAC_MAX_DATA_TX_RETRIES (3U)|
Maximum TX transmission retries for DATA packet in case of no response from the receiver.
When a data packet is scheduled for transmission, i.e., pushed into TX for sending, LWMAC defines a maximum of GNRC_LWMAC_MAX_DATA_TX_RETRIES retries for transmission of the packet. That is, in case of transmission failure in TX due to no WA from the receiver, the sender will not drop the packet, but keeps it and retries to send the data packet in the following cycles, until the sender reaches the maximum retries limit defined here. Then, the packet will be dropped.
|#define GNRC_LWMAC_MAX_RX_EXTENSION_NUM (3U)|
MAX bad Listen period extensions a node can tolerate.
In LWMAC, to allow burst transmissions, when in the wake-up period and by default, a node will extend its wake-up period to another GNRC_LWMAC_WAKEUP_DURATION_US after each packet reception (except for broadcast packet). However, in some cases, a receiver may overhear other unintended packets, e.g., WR or WA packets for other nodes, these are called bad extensions for the receiver. If a receiver reaches the maximum bad listen extension limit defined here, it goes to sleep mode with the consideration that the channel is currently unavailable/busy.
|#define GNRC_LWMAC_MAX_TX_BURST_PKT_NUM (GNRC_LWMAC_WAKEUP_INTERVAL_US / GNRC_LWMAC_WAKEUP_DURATION_US)|
MAX burst transmission packet number in one shot.
LWMAC supports burst transmission based on the pending-bit technique, and this macro here defines the largest number of packets allowed to be sent in one consecutive sequence. In case a sender has multi packets for one receiver,the burst transmission procedure is as follow:
|#define GNRC_LWMAC_PREAMBLE_DURATION_US ((13LU * GNRC_LWMAC_WAKEUP_INTERVAL_US) / 10)|
The Maximum WR (preamble packet gnrc_lwmac_frame_wr_t) duration time.
Since LWMAC adopts duty-cycle scheme, a node only wakes up for a short period in each cycle. Thus, to probe where is the wake-up period of the receiver, a sender sends WR (preamble) packets to notice the receiver for communication. To ensure that the receiver will catch at least one WR packet in one cycle, the sender repeatedly broadcasts a stream of WR packets with the broadcast duration (preamble duration) slightly longer period than GNRC_LWMAC_WAKEUP_INTERVAL_US.
|#define GNRC_LWMAC_TIME_BETWEEN_BROADCAST_US (GNRC_LWMAC_TIME_BETWEEN_WR_US)|
Time to idle between two successive broadcast packets, referenced to the start of the packet.
The same limitation as for GNRC_LWMAC_TIME_BETWEEN_WR_US apply here. In LWMAC, when a sender initiates a broadcast, it starts with sending a stream of repeated broadcast packets with GNRC_LWMAC_TIME_BETWEEN_BROADCAST_US interval between two consecutive broadcast packets. After sending one broadcast packet, the sender turns to the listen mode with a timeout of GNRC_LWMAC_TIME_BETWEEN_BROADCAST_US. When this timeout expires, the sender sends the next broadcast packet until reaching the maximum broadcast duration of GNRC_LWMAC_BROADCAST_DURATION_US.
|#define GNRC_LWMAC_TIME_BETWEEN_WR_US (5U * US_PER_MS)|
Timeout to send the next WR in case no WA has been received during that time.
In LWMAC, when a sender initiates a transmission to a receiver, it starts with sending a stream of repeated WR packets with GNRC_LWMAC_TIME_BETWEEN_WR_US interval between two consecutive WRs. After sending one WR (preamble) packet, the sender turns to the listen mode to receive the potential incoming WA (preamble-ACK) packet with a timeout of GNRC_LWMAC_TIME_BETWEEN_WR_US. If no WA is received during GNRC_LWMAC_TIME_BETWEEN_WR_US, the sender starts sending the next WR. It is referenced to the beginning of both WRs, but due to internal overhead, the exact spacing is slightly higher. The minimum possible value depends on the time it takes to completely send a WR with the given hardware (including processor) and data rate.
|#define GNRC_LWMAC_WAKEUP_DURATION_US (GNRC_LWMAC_TIME_BETWEEN_WR_US * 2)|
How long a node in LWMAC should keep awake and listen on the channel in one cycle.
LWMAC adopts the duty-cycle scheme that a node only wakes up for a short period of GNRC_LWMAC_WAKEUP_DURATION_US in each cycle. In the rest of the cycle, the node turns off the radio to conserve power. GNRC_LWMAC_WAKEUP_DURATION_US is set to twice the duration of GNRC_LWMAC_TIME_BETWEEN_WR_US, to guarantee that the wake-up period is long enough that receiver will not miss the WR (preamble) packet. Receiver needs to support NETDEV_EVENT_RX_STARTED event in order to use time-between-WR as a sensible default here. Otherwise the duration of WRs as well as longest possible data broadcasts need to be taken into account.
|#define GNRC_LWMAC_WAKEUP_INTERVAL_US (100LU * US_PER_MS)|
Time between consecutive wake-ups.
This macro governs power consumption, latency and throughput! In LWMAC, devices adopt duty-cycle scheme to conserve power. That is, time is divided into repeated cycles (or, superframes), and in each cycle, a node only wakes up for a period of time for receiving potential incoming packets for itself. This macro defines the wake-up interval, or, in other words, defines the cycle duration used in LWMAC. If the wake-up interval is short, nodes will wake up more frequently, which also increases the chances for receiving packets from neighbors (i.e., leads to higher throughput), but also results in higher power consumption. In LWMAC, by default, we regard the wake-up period as the beginning of a cycle.
|#define GNRC_LWMAC_WR_PREPARATION_US ((3U * US_PER_MS))|
WR preparation overhead before it can be sent (higher with debugging output).
In LWMAC, when a sender wants to send a data packet to the receiver, it starts sending the WR stream a little bit earlier (advance) to the beginning edge of destination's wake-up phase over time. The idea is not to miss the wake-up period of the receiver, otherwise will lead to a long WR procedure.
|kernel_pid_t gnrc_lwmac_init||(||char *||stack,|
|const char *||name,|
Initialize an instance of the LWMAC layer.
The initialization starts a new thread that connects to the given netdev device and starts a link layer event loop.
|[in]||stack||stack for the control thread|
|[in]||stacksize||size of stack|
|[in]||priority||priority for the thread housing the LWMAC instance|
|[in]||name||name of the thread housing the LWMAC instance|
|[in]||dev||netdev device, needs to be already initialized|